The persistent educational differences in mortality in the United States represent a troubling and challenging issue. Recent research finds that U.S. adults with less than a high school education had nearly twice the risk of mortality in a five-year follow-up period compared to adults with graduate degrees. This translates into 5-to-7 years of differential life expectancy at age 25 across educational groups, depending on the specific sex and racial/ethnic group in question. The overall goal of our research is to improve the understanding of the linkage between educational attainment and overall and cause-specific adult mortality within the population as a whole and among various subgroups of the adult population. Although the overall relationship between [unreadable] educational level and adult mortality risk has been quite well documented since the classic Kitagawa-Hauser (1973) study that used 1960 data, surprisingly little attention has been given to how this relationship varies by race/ethnicity, nativity, gender, age, and cause of death. Further, there are considerable debates regarding the extent to which educational differences in mortality have been widening or narrowing over time, whether or not educational differences in mortality widen or narrow with increasing age, and how other socioeconomic and health variables mediate the relationship between education and mortality risk. We will use data from the National Health Interview Survey-Multiple Cause of Death (NHIS-MCD) linked files to accomplish our goal. The MCD links the National Death Index through 2002 to the 1986 through 2000 cross-sectional waves of the NHIS. These linked data provide information on: (1) nearly 1 million U.S. adults aged 25 and above followed for subsequent survival status for up to 17 years; (2) over 100,000 identified subsequent deaths, with information on both date and cause of death; (3) enough racial/ethnic/nativity, gender, and age diversity to examine very detailed patterns of mortality for specific subgroups of the population; (4) a long enough time period to examine duration and cohort differences in the education mortality relationship. Further, measures of socioeconomic status (including family income, employment status, and occupational status) and information on cigarette smoking and body weight and height are available in these data, which will allow us to better understand how education, at least in part, may work through other socioeconomic variables, cigarette smoking, and weight-for-height to influence adult mortality risks in the United States. The proposed research addresses a topic of immense scientific and public policy [unreadable] interest and will help to inform a great deal of other current work that examines educational differences in health changes, health behavior, and disability trajectories throughout the life course. [unreadable] [unreadable] [unreadable] [unreadable]